DQ4 Database Selection Criteria | LCA Database Quality Rapid Assessment Framework

by Raymond Wang · SFiTCorp.com and SSBTi.org 2026

‍【Summary】 DQ4 is a four-question rapid assessment framework for determining whether an LCA database is "usable and trustworthy." Originating from Bill Kung's Beyond the Trap: Product Carbon Footprint and Value Synergy — used as course material in SFiT Corp and SSBTi's joint Scope 3 training with 1mi1, and subsequently refined and adapted — the framework serves as the selection criteria for Nanozeo.com.tw as an authorized distributor and database screening partner, and as the decision-making methodology for SFiT Corp / SSBTi in evaluating database adoption. This article also includes practical application guidance. SFiT Corp uses this as its database-selection methodology foundation, helping enterprises make the right choice before adopting an LCA database — avoiding "buying the wrong database" or "buying one they can't use."

In one sentence: DQ4 isn't about finding the perfect database — it's about filtering out the ones you can't trust.

The Four DQ4 Questions

Q1 | Independent Review

Has this dataset been reviewed by an independent professional to verify compliance with ISO 14040 / 14044?

SFiT Practical Interpretation: Has the database or its unit process data been reviewed by an independent third party (not the database builder themselves)? Does the review scope cover LCA methodological compliance? This is the first gate for determining whether data is "acceptable to third-party verification."

Indicators:

• Does the database's official documentation list the reviewing body and review scope?

• Can an independent review report be produced (e.g., ILCD-EL Review Report)?

• Note: Builder's "self-declared compliance" ≠ independent review

• 
  

Q2 | Data Accessibility and Verification

Are the raw material, auxiliary material, energy consumption, and emission data publicly accessible, and has the data collector performed basic data verification — such as mass balance analysis, data quality analysis, reliability testing, anomaly detection, uncertainty analysis, or other documentation?

SFiT Practical Interpretation: A database cannot be a "black box" — giving you only a carbon factor without explaining how it was derived. A credible database must at minimum allow traceability: what materials were used, how much energy was consumed, what was emitted, and whether the data underwent mass balance verification.

Indicators:

• Does it provide unit process-level data, not just system process-level data?

• Is there a record of mass balance verification?

• Is there uncertainty analysis (e.g., Monte Carlo simulation results)?

• ecoinvent is the benchmark here (provides unit process data + uncertainty distributions)

• 
  

Q3 | Source Transparency and Representativeness

Can the cited literature or sources adequately demonstrate that the unit process data meets ISO 14040 / 14044 analytical requirements, and is there complete, transparent information on the data's temporal, technological, geographical, and precision characteristics for user reference?

SFiT Practical Interpretation: This is the core of DQI (Data Quality Indicator) — temporal representativeness (TiR), technological representativeness (TeR), geographical representativeness (GR), and precision (P). A dataset that cannot even specify "which year, which country, which technology" cannot be used in formal LCA modeling.

Indicators:

• Is there a clear timestamp (data vintage; generally, data >10 years old is not accepted)?

• Is the technology type labeled (e.g., "China grid average" vs. "plant-specific measured data")?

• Is the geographical scope labeled (global / regional / national / site-level)?

• Is there a sample size and precision statement?

• If local data comes from outdated processes, it may be better to use international data of equivalent advanced technology with localization adjustments

  
  

Q4 | Certification Endorsement

Does the data provider have sufficient grounds to demonstrate that the data meets ISO 14040 / 14044 modeling and data requirements, with adequate assurance — for example, the database has supported product assessments that have passed rigorous application or certification audits where the audit scope included the database, but details cannot be publicly disclosed due to confidentiality?

SFiT Practical Interpretation: Some databases cannot fully disclose unit process details due to commercial confidentiality. However, if their data has been used in assessments that passed rigorous third-party audits (EPD, CBAM, or SBTi), this serves as indirect endorsement. This is the final "fallback path" — if a database fails the first three questions but has been battle-tested, it can still be trusted.

Indicators:

• Has this database supported published EPD reports that passed audit?

• Is the auditing body credible (e.g., TÜV, SGS, EPD program-accredited reviewers)?

• Does the audit scope explicitly cover the background database?

Core Rule

<aside>🎯

If a database passes ANY ONE of the four DQ4 questions, it can be considered acceptable for reference and public use based on applicability.

DQ4 is designed as a threshold-based framework, not a scoring system — it doesn't require passing all four, just that at least one path is viable. The pragmatic reasoning: no database is perfect, but it must withstand at least one form of scrutiny.

SFiT Application: DQ4 Rapid Database Screening Card

This is the selection workflow SFiT Corp uses when helping clients integrate LCA databases:

Step	Action	Pass Criterion
Step 1	Request independent review report from the database provider	Q1 ✓
Step 2	Check for unit process-level data + mass balance verification	Q2 ✓
Step 3	Confirm TiR / TeR / GR / P representativeness info is complete	Q3 ✓
Step 4	Query whether the database has supported published EPD / CBAM audits	Q4 ✓
Decision	Any one question passed → Usable; all four failed → Not recommended	—

DQ4 Quick-Screening Matrix

Common Database	Q1 Independent Review	Q2 Data Accessible	Q3 Source Transparent	Q4 Battle-Tested	Overall
ecoinvent	✓	✓ (unit process)	✓	✓	✅ Full path
CLCD / CPCD	Partial ✓	Partial	✓ (Asia-local)	✓ (China market)	✅ Asia scenarios
GaBi (Sphera)	✓	Partial (software-locked)	✓	✓	✅ Sphera users only
ELCD / PEF	✓	System process only	Partial	✓	⚠️ Insufficient detail
AI-generated / web-scraped databases	✗	✗	✗	✗	❌ Not recommended

DQ4 vs. DQR (ILCD Data Quality Rating)

DQ4 is the threshold — "can it be used?"; DQR is the rating — "how good is it?"

• DQ4: Screening phase — quickly determine whether a database is worth importing

• DQR: Modeling phase — for each dataset in the selected database, calculate the weighted rating across six indicators (TeR / GR / TiR / C / P / M), ensuring it meets at least Basic quality (DQR ≤ 3)

The two complement each other: DQ4 guards the entry gate; DQR guards game time.

Common Pitfalls and SFiT Warnings

Pitfall 1: Assuming "more data" = "better quality"

Many new databases boast hundreds of thousands of data entries, but upon closer inspection, they are fragmented — no industry chain or life-cycle stage relationships. That's not an LCA database; it's just a collection of data sources.

Pitfall 2: IPCC default values ≠ LCA database

A "carbon database" that provides only a single carbon factor (e.g., IPCC GWP values) cannot support multi-indicator calculations (PEF requires 16+ indicators). Buying such a database means buying a tool that can only calculate carbon — and nothing else.

Pitfall 3: AI-generated databases

Databases auto-generated via web scraping or ChatGPT / DeepSeek have no enterprise authorization, no data verification, and no methodological compliance — "you cannot build a carbon footprint edifice on a foundation of quicksand."

Pitfall 4: The local-data fallacy

"Local data must be better" is wrong. If local data comes from outdated processes (>10 years old, small sample size, remote areas), it is better to use international data of equivalent technology (e.g., from ecoinvent) with localization adjustments and uncertainty analysis — this is SFiT's service strength.

SFiT Corp's Service Positioning

<aside>💡

SFiT Corp is the "database consulting team" that handles DQ4 screening + DQR modeling oversight.

We don't compete with carbon management consultants. We simply clear the path at the database layer — helping you select the right database, build the right model, and pass audits — so your consulting team can work from a clean, compliance-ready data foundation.

</aside>

• Database and software license procurement → Nanozeo.com.tw

• DQ4 database selection and compliance integration → sfitcorp.com

• 
  

Source: Bill Kung, Beyond the Trap: Product Carbon Footprint and Value Synergy*, Chapter 2 (print pp. 107–121). Used as course material in SFiT Corp and SSBTi's joint Scope 3 training with 1mi1. Adapted and refined by SFiT Corp as the selection criteria for Nanozeo.com.tw as an authorized distributor and database screening partner, and as the decision-making methodology for SFiT Corp / SSBTi in evaluating database adoption. This article also includes practical application guidance. Original DQ4 question text copyright remains with the author.*